Drug Development Research 13:169-185 (1988)
Percutaneous Drug Penetration: Choosing
Candidates for Transdermal Development
Gordon L. Flynn and Barbra Stewart
College of Pharmacy, Pharmaceutics Group, College of Pharmacy,
The University of Michigan, Ann Arbor
ABSTRACT
Flynn, G.L., and B. Stewart: Percutaneous drug penetration: Choosing candidates for
transdermal drug development. Drug Dev. Res. 13:169-185, 1988.
There is currently a high level of interest in using the skin as a route for delivering drugs.
One hears the questions: What are the attributes of a drug that make it a serious
candidate for transdermal delivery? By what a priori analysis might one zero in on the
best transdermal candidate within a family of drugs? Answers to these questions lie in
understanding the molecular factors that make a drug a facile permeant of the skin.
Among other properties, it must have a high absolute affinity for the skin’s phases, which
provide for its diffusive conduction. Other factors in evaluation are the potency of the drug
and the relative efficiency of the drug’s systemic presentation once it has gained access
to the body. One also considers the potential for the drug to elicit adverse responses in
the skin. Fortunately, parallels between the drug’s ability to partition between oil and
water and its ease of mass transfer across the skin can be used to ferret out a working
mass transfer coefficient. If not already known, solubilities are easily experimentally
deduced. The extent of first-pass metabolism by the oral route, presumed to be a known
quantity, is compared with the relative amount of metabolism of the drug in the course of
its diffusion through the skin, an experimentally determined quantity, in order to set the
transdermal dose. These bits of information can then be used to form an early, reasonably
faithful picture of the feasibility of delivering a particular drug transdermally and to make
a first estimate of the size of patch required for the drug.
Key wards: skin permeability, transdermal delivery, percutaneous absorption, structureactivity relationships, selecting drug candidates
Received final version January 26, 1988; accepted Janaury 28, 1988.
Address reprint requests to Dr. Gordon L. Flynn, College of Pharmacy, The University of Michigan, Ann
Arbor. MI 48109-1065.
0 1988 Alan
R. Liss, Inc.
170
Flynn and Stewart
INTRODUCTION
For as long as it has been appreciated that substances exist in nature that alleviate pain
and suffering, man has sought the means to administer them to gain their full advantage. One
can presume that the oral route figured prominently in prehistoric medicine, for taking drugs
orally is as natural as eating. One can also presume that early man encountered some of the
general difficulties that still exist with the oral route, namely problems attending administration of drugs to nauseous and unconscious patients. Alternative methods of administration
were surely attempted, with the topical route certainly amongst those tried, given its
accessibility and extensiveness. Drugs were apparently used topically by early Egyptian and
other Mediterranean civilizations not only to treat surface wounds and the local manifestations
of disease but also to cure general illnesses. Poultices, cataplasms, and plasters were still in
wide use as cold remedies and for other human afflictions at the dawn of the 20th century.
However, only in the last 20 years has it proven possible to unequivocally demonstrate that
systemic therapy by the topical route, though subject to severe constraints, is indeed feasible.
We seem to be passing through a period in which the commercial lure of transdermal
delivery is running out in front of our comprehension of the administration method. Many,
many hours have therefore been spent in the laboratory and clinic learning the hard way that
many drugs fail stringent tests of potency and toxicity requisite for their delivery through the
skin. Wouldn’t it be nice if screening to sort out the most promising drugs for the method could
be accomplished in the library with the aid of pencil and paper and a keen sense of the
processes involved? If one can accept an order of magnitude answer, this is, in fact, possible!
The present report aims to establish this point. It catalogs obstacles to the successful
deployment of the transdermal method of delivery and especially concentrates on the
physicochemical and biophysical constraints associated with permeation of the skin. The
relatively impermeable strata of skin interposed between the surface of the body and the
systemic circulation are a prime consideration. Although a conservative stance is taken with
respect to the delivery method, this report actually places transdermal delivery in a positive
light, for, when the principles set forth are applied, one can rule out unworthy drugs and
concentrate effort on candidates of high promise. It emphasizes two points: 1) the information
required for the initial judgement about the transdermal potentials of drugs; and 2) the
application of the information to make enlightened decisions. While the topical delivery of
drugs to achieve their systemic effects is of sole concern, the general approach to the selection
of drug candidates can also be applied to the selection of drugs best suited for local therapy.
PERCUTANEOUS ABSORPTION
Transepidermal Absorption
Percutaneous absorption involves passive diffusion of substances through the skin. The
mechanism of permeation can involve passage through the epidermis itself or diffusion through
shunts, particularly those offered by the relatively ubiquitously distributed hair follicles and
eccrine glands (see Fig. I). While exceptions to the rule are acknowledged, it is now generally
believed that the transepidermal pathway is principally responsible for diffusion across the
skin. Far more often than not, the main resistance encountered along this pathway arises in the
stratum corneum. Characterization of the physical nature of the skin barrier thus should begin
with characterization of the stratum corneum.
Over most of the body, the stratum corneum is composed of 15 to 25 layers of “acutely
flattened, metabolically inactive cells” [Odland, 19831. It is surprisingly dense for a soft
tissue, having an estimated dry weight density of 1.3-1.4 gm/cm3 [Scheuplein, 1978;
Scheuplein and Bronaugh, 19831. The thickness of individual cell layers varies, ranging from
0.2 pm to 0.5 km, depending on location [Plewig et al., 19831, but appears to be uniform at
a location. The individual platelets, which function as building blocks, appear as crudely
Percutaneous Drug Penetration
171
TRANSEPIDERMAL
TRANSECCRINE
I
Fig. 1 . Illustration showing the possible routes for percutaneous absorption. Currently it is believed that
a transepidermal route through the seams between the “cells” of the stratum corneum is the principal
pathway for most drugs.
shaped polygons. In their longest dimension these measure 30-40 Fm. By the time the cell
platelets are laid into the stratum corneum, the cell interior is crisscrossed with densely packed
bundles of keratin fibers [Steinert, 19831. Due to this, the dry composition of the horny layer
is 75-85% protein, most of which is intracellular keratin. A fraction of the protein also appears
to be associated with a still intact network of cell membranes. The bulk of the remainder of the
substance of the stratum corneum is lipids. Elias and coworkers [ 19831 make a strong case that
the stratum corneum’s lipoidal substance lies almost exclusively in the seams between cells.
It appears to be organized into bilayers. The greater part of this lipid originates from membrane
coating granules found in the granular layer. These pass from the granular cell’s interior into
the intercellular space as one of the culminating events of differentiation of the epidermal
keratinocyte. The stratum corneum thus has two distinct chemical regimes, the mass of
intracellular protein and the intercellular lipoidal medium. These phases are isolated from one
another by still present, and presumably intact, cell membranes. The cell membranes
themselves are knit together by desmosomes, adding a tough infrastructure to the horny mass
[Goldsmith, 19831. This conceived structural organization supports the idea that a lipoidal
route exists between the cells, possibly with alternative pathways through the mass of protein
[Scheuplein, 1965, 1967, 1976, 1978; Scheuplein and Bronaugh, 1983; Katz and Poulsen,
1971; Flynn, 19851. The intercellular lipid route occupies 1% of the stratum corneum’s
productive diffusional area (area in the plane of the skin and perpendicular to permeation) by
rough estimate [Flynn, 19851. As meager as it is, this description captures most of what is
accepted about the organization of the stratum corneum’s physically distinct phases. The
picture arrived at is to an uncomfortable degree still tentative, making a more refined
description of the tissue’s potential diffusion pathways hazardous.
172
Flynn and Stewart
When a permeating drug exits the stratum corneum, it enters the wet cell mass of the
epidermis, and, since the epidermis has no direct blood supply, the drug is forced to diffuse
across it to reach the vasculature immediately beneath. Given that this mass of viable cells
gives rise to the stratum corneum, it is remarkable how different its barrier properties are. In
the live epidermis several histologically distinguishable layers of closely packed cells form a
tight, cellular continuum 50-100 pm in thickness. At its base, the epidermis forms an
irregular, papillate interface with the dermis. The basal cells are overlaid on a thin basement
membrane and are anchored to the dermis by tiny fibers. Proliferating stem cells in the basal
layer give rise to all epidermal cell layers above. The basement cells tend to be cuboidal or
columnar in shape, but, as they progress toward the horny layer, they undergo profound
changes in appearance. Sharp, spiny protuberances form at points of cell attachment
immediately above the basal layer (stratum spinosum). As the cells near the undersurface of
the stratum corneum, they flatten considerably and stain to give a granular appearance (stratum
granulosum). The nucleus and cell organelles are still visible in the granular cell’s interior as
are keratohyalin substance, the preferential staining of which gives the granular appearance,
and small lipid-filled vesicles known as membrane coating granules. The latter particles pass
by exocytosis into the space between cells in the final stage of formation of the stratum
corneum. Though appearing granular and full of vesicles and cell organelles, there is little
reason to believe that the barrier qualities of the granular layer are unique within the viable
epidermis. Rather, and as a practical matter, the viable epidermis, mitotic layer to granular
layer, is considered as a single field of diffusion in models. A considerable amount of
experimental and other evidence indicates it is a permeable field that functions as a viscid
watery regime to most permeants [Scheuplein, 1965, 1967, 1976; Flynn, 19851. It appears that
only ions and polar nonelectrolytes found at the hydrophilic extreme and lipophilic nonelectrolytes at the hydrophobic extreme have any real difficulty passing through the viable field
[Flynn, 19851. The epidermal cell membranes are tightly joined and there is little to no
intercellular space for ions and polar nonelectrolyte molecules to diffusionally squeeze
through Thus permeation must follow a course that requires frequent crossings of cell
membranes, each crossing being a thermodynamically prohibitive event for such water-soluble
species. Extremely lipophilic molecules, on the other hand, are thermodynamically constrained from dissolving in the watery regime of the cell (the cell cytoplasm).
A rich bed of capillaries is encountered 20 pm or so into the dermal field. Passage
through this dermal region represents a final hurdle to systemic entry. This is so regardless of
whether permeation is transepidermal or by a shunt route. The dermal construction is a mesh
of collagen fibers and is the principal structural element of skin. Wide spaces between the
bundles of collagen fibers are filled with a watery gel called the ground substance. Permeation
through the dermis is through the interlocking channels of ground substance formed between
fibers. The gaps between fibers are far too wide to filter large molecules and thus, as far as can
be told from existing information, diffusion through the dermis is facile and without molecular
selectivity. Because the viable epidermis and dermis lack major physicochemical distinction,
they are generally considered as a single field of diffusion [Flynn, 19851. An exception to this
generalization is made for solutes of extreme polarity, as the epidermis offers a factorable
resistance to such species.
Permeation by the transepidermal route first involves partitioning into the stratum
corneum. Diffusion across this tissue, the viable epidermis, and the upper dermis then ensues.
The current popular belief is that most substances diffuse across the stratum corneum via a
lipoidal pathway, which recent histological evidence places between the cells [Elias et al.,
1983). This is a tortuous pathway of limited fractional volume and even more limited
productive fractional area in the plane of diffusion. The presence and nature of this pathway
explain the greater part of the existing chemical structure-skin permeation data [Flynn, 19851.
However, there appears to be another microscopic path through the stratum corneum for
extremely polar compounds and ions. Otherwise these would not permeate at rates that are
Percutaneous Drug Penetration
173
measurable, considering their o/w distributioning tendencies [Flynn, 19851. In the other
extreme of polarity, lipophilic molecules concentrate in and diffuse with relative facility
through the horny layer’s intercellular regime only to find themselves blocked by the wet cell
mass beneath. Thus the viable tissue is rate-determining when nonpolar compounds are
involved [Scheuplein, 1965, 1967, 1976; Flynn, 19851.
Transfollicular (Shunt Pathway) Absorption
Arguments have raged about the true pathway for percutaneous absorption, but it is now
the generally accepted view that the skin’s appendages offer only secondary avenues for the
permeation of most chemicals. Sincc pilosebaceous and eccrine glands are distributed over the
entire body, except for some very limited, isolated areas, only these appendages are ever taken
seriously as shunts bypassing thc stratum corneum. Though eccrine glands are numerous, up
to 400 glands per cm2 in places such as the palms and soles, their orifices are tiny and add up
to a miniscule fraction of the body’s surface. Moreover, they are either evacuated or so
profusely active that it is unlikely that molecules can diffuse inwardly against the gland’s
output. For these reasons they are discounted as a significant route for percutaneous
absorption.
Pilosebaceous glands are less numerous than eccrine glands, having a density variously
estimated to be between 50 and 100 glands per cm2 of body surface. However, the opening of
the follicular pore where the hair shaft exits the skin is relatively large; the fractional area of
the route has been estimated to be as high as I/l,000th of the skin’s surface [Flynn, 19851.
Moreover, the duct of the gland is filled with a soft, slowly extruded lipoidal medium, sebum,
in which lipophilic drugs should be soluble and through which they should diffuse with ease.
Thus the follicular route remains as a possibly important avenue for percutaneous absorption.
Partitioning into sebum followed by diffusion through the sebum to the depths of the epidermis
is the envisioned mechanism of permeation by this route. Vasculature subserving the hair
follicle located in the dermis represents the likely point of systemic entry,
Clearance by the Local Circulation
The earliest possible point of entry of drugs and chemicals into the systemic circulation
is within the papillary plexus in the upper dermis. The process of percutaneous absorption is
generally regarded as ending at this point, a presumption that is almost certainly a model-based
oversimplification, as some molecules clearly bypass the circulation and diffuse deeper into
the dermis. Blood flow through the superficial capillary plexus is highly variable. For one
thing, the level of perfusion of blood is determined by the existing physiological need to
conserve or eliminate body heat. Whether the flow of blood through chilled skin is adequate
for efficient local clearance of pcrmeant is an important question having no current answer.
Blood flow is also exaggeratedly increased by local trauma. On the other hand, clearance may
be diminished in old age due to vascular decrepitude. This thought has sparked the suggestion
that aged skin is more prone to the allergic and irritant effects of topically contacted
compounds [Kligman, 19791. Despite these collective concerns, the preponderance of
evidence seems to indicate that local clearance is relatively efficient.
The above is a thumbnail skctch of the principal features of the skin as they relate to the
inward diffusion of chemicals from the skin’s surface. It is clear that the development of
models for skin permeation must include diffusion across a series of barriers and possibly
diffusion through parallel pathways. Moreover, certain anatomical structures of the skin are
relatively impervious, a factor that limits the available area for diffusion and forces permeation
mainly through the limited fraction of soft lipoidal material between the stratum corneum cells
or through sebum in the follicular duct. Clearance is by way of the blood, the flow of which
varies depending on momentary physiological need.
174
Flynn and Stewart
SKIN PERMEATION ASSESSMENT
Diffusion From a Saturated Medium in Contact With the Skin
In order to make an assessment of the systemic availability of a topically applied drug
(or topically contacted chemical), one normally has to consider the separate but concerted
delivery processes of drug release from the topical vehicle (or medium containing the
chemical) into the skin and diffusion through the skin. However, to assess whether a drug can
be delivered transdermally for systemic therapy, it is only necessary to consider the latter
process. This is so because it is presumed ah initio that the means will be found to maximize
(or optimize) a drug’s thermodynamic potential at the skin’s surface, providing the drug’s
percutaneous absorption rate is consistent with therapeutic needs. Removing all considerations
of vehicle control to make the first assessment of feasibility of delivery considerably simplifies
the analysis. What one in effect does is estimate the maximum rate at which the drug can be
caused to diffuse across the skin. It’s easy to back off from this rate should it be overly fast.
The maximum rate will be achieved at the attainable limit of thermodynamic activity of the
drug. For a crystalline drug (most drugs of interest are going to be in the form of molecular
crystals), the maximum attainable thermodynamic activity is set by the neat solid. Therefore,
for the purpose of establishing feasibility, transdermal drug delivery from a continuously
saturated solution through the skin is assumed. While the dominant resistance of the skin
usually lies in its stratum corneum, the outermost, 10-Fm-thick, devitalized layer of the tissue,
there are times and circumstances when other strata of the skin in series with the stratum
corneum offer appreciable resistances to diffusion or when shunt routes are followed and
cognizance is taken of this in the general analysis.
Situations in which a drug is delivered from a reservoir in which it is at unit activity
(from a saturated solution) through a resistant membrane like the stratum corneum (or whole
skin) are frequently encountered in transdermal delivery. Equation 1 mathematically describes
the delivery dependency:
Here M, is the amount of a drug passing through a unit area of the membrane in time, t, under
conditions that the drug is delivered from a saturated reservoir of concentration, C,. K is the
skidvehicle partition coefficient, making the product, K C , , the saturation concentration of the
drug within the skin’s surface. The terms within the summation, D,, and h,,, are the efsective
diffusion coefficient and the effective thickness of the stratum corneum or, more generally, the
membrane, respectively. Though subscripted with sc to designate the stratum corneum, they
are in fact functional parameters representing the net properties of all consequential strata.
Even when the stratum corneum is the sole source of diffusional resistance, D,, and h,, are
eflective parameters because there is no way to make an explicit accounting for important
factors such as the fractional area and tortuosity of the route taken through the stratum
corneum.
The theoretical curve (schematically drawn) depicted in Figure 2 is based on equation 1.
The curve indicates that no drug or very little drug penetrates the skin in the early moments
after the application of the drug-containing reservoir. However, once the first molecules
breach the membrane, the flux of the drug gradually grows until it reaches a steady level. One
obtains the equation for the steady-state line by allowing t + 00, which leads to:
175
FI
w
CI
c
a
CI
w
2
w
pc
E
z
3
0
B
4
w
+-
n
CI
4
GI
3
5i
3
u
TIME
Fig. 2. Schematic representation of the standard permeation profile (amount penecrated through the
membrane as a function of time), which is obtained using a two-compartment diffusion cell. The intercept
of the steady-state line (equation 2 in the text) is the lag time (equation 4 in the text).
M,
=
[
K . D,, . C, t - - h s 2 )
hsc
6D,,
In effect, equation 2 describes a situation in which the amount penetrating per unit time is
constant, the steady-state condition. The steady-state flux, dM,/dt, is readily obtained from
equation 2 by differentiating M, with respect to t, i.e.:
dM,
-
dt
=
K
. D,, . C,
(3)
h,,
When the steady-state line is extrapolated to the time axis, the value o f t at M
to be:
=
0 is found
This intercept, tL, is refcrred to as the lag time. In essence, it is a measure of the time it takes
for the permeant’s concentration gradient to become stabilized across the membrane. In
Flynn and Stewart
176
principle (with the simplest of membranes), the diffusivity can be calculated from tL providing
the membrane thickness is known. With a complex membrane such as the skin, only an
eflective value of the diffusivity is obtained.
The essential meanings of equations 1 through 4 can be considered. Equation 4 tells us
that the lag time for penetration becomes longer as the effective diffusivity, D,,, becomes
smaller (irrespective of how it is calculated). Normally one wants the lag time to be a
negligible fraction of the projected time of use of a transdermal system. If the prescribed period
of application is to be a day, as it is with current nitroglycerine patches, the lag time should
be no more than a tenth of a day or 2-3 hours at the most. Any lag time longer than this raises
questions about the utility of the patch, for the patch will not be delivering drug over a
significant part of the time it is in use. Consequently, the attainment and maintenance of even
blood levels will be next to impossible. In vitro diffusion cell studies suggest that the values
of D,, for the stratum corneum (one uses the stratum corneum’s 10-km thickness in the
calculation) range from lop9 down to
cm2/sec [Scheuplein, 1978; Scheuplein and
Bronaugh, 19831. At the larger diffusivities, lag times of several minutes are expected. On the
other hand, multiple-day lag times are suggested by the smallest reported diffusivities. In our
actual experience, diffusional lag times range from several hours to over half a day for
nonelectrolyte drugs having molecular weights of 200 daltons or more. Moreover, the more
polyfunctional and polar a compound is, the longer its lag time seems to be. For example, lag
times across cadaver skin membranes for the permeation of fentanyl and sufentanil, both
narcotics of low polarity based on partitioning data, have consistently been less than 3 hours.
In contrast, lag times for morphine and hydromorphone, both relatively polar narcotics, have
consistently exceeded 6 hours, well over than twice those of fentanyl and sufentanil. This
alone poses a problem with respect to their transdermal delivery. Clearly, thefirst obstacle to
transdermal delivery is overly long lag times. Fortunately, in vitro diffusion cell methods
appear to be able to indicate when this is going to be a problem.
Equation 1 indicates that, providing enough time is allowed to elapse, a steady state in
permeation is obtained. The rate of delivery in the steady state is described by equation 3. The
steady state arises in part because delivery is from a reservoir of constant thermodynamic
activity. The development of a steady state also requires the barrier properties of the skin
membrane to be invariant over the lifetime of patch use (or over the duration of a diffusion cell
experiment). For the purpose of analysis, we will assume the skin membrane is stable, in
which instance equation 3 tells us that the per unit area rate of delivery is proportional to the
product of the saturation concentration, C,, diffusivity, D,, and the partition coefficient, K ,
as well as inversely proportional to the thickness of the membrane, hsc. The product, K . C,,
yields the saturation concentration in the surfacemost structure of the stratum corneuml and is,
at least theoretically, independent of the nature of the medium used in the reservoir. The
process is driven by the activity of the solid state; the gradient across the skin directly reflects
this activity. For a complex membrane such as the skin, derived values for D,,, K , and h,, have
magnitudes that depend on assumptions about the nature of the membrane. Since they cannot
be experimentally disentangled, they are commonly lumped together in a single term. One thus
arrives at:
dMt = P C,
+
=
constant
dt
‘A mechanism of permeation of a drug involving diffusion of the drug across the stratum corneum, the
so-called transepidermal pathway, is assumed. However, the principles set forth arc general with the
assumption having no practical consequence on the analysis.
Percutaneous Drug Penetration
177
where P is the mass transfer coefficient or “permeability coefficient. ” The total flux in the
steady state is therefore:
where now the area, A, is explicitly considered. It is obvious that, in any designated small
increment of time, the total amount of drug that permeates the membrane, dMt,Totd,is equal
to the volume of the collecting phase, V, times the change in concentration of this phase, dC,.
It is these forms of equation 3 that are used to calculate permeability coefficients from data
gathered in in vitro studies. This is possible because the area available for diffusion in the
diffusion cell and the solubility or, more generally, concentration of the drug in the medium
applied to the skin mounted in the diffusion cell are independently measured and known prior
to starting an experiment. The in vitro experiment is begun, and the flux in the steady state is
measured. From these, one calculates the proportionality constant between flux and the
product of area and concentration. This is what the permeability coefficient actually is. It takes
a value that is uniquely dependent on: 1) the chemical structure of the drug; 2) the
physicochemical nature of the medium of application; and, of course, 3) the physicochemical
nature of the membrane.
When the membrane and medium of application are fixed, differences in permeability
coefficients through a family of drugs are strictly related to the differences in chemical
structures of the drugs. Clearly, the larger the value of P, the greater the flux of the compound,
all other factors being equal. According to Flynn 119851, permeability coefficients for hairless
mouse skin determined from saline (or isotonic aqueous buffer) range from about
to
about
cmhour, with indications that they take levels ten times smaller in human
epidermis on the lower side of the range. Using 1 mgiml as a reference concentration2 and the
maximum permeability coefficient of lo-* cmihour, one calculates that 20 pg of drug passes
through the skin membrane pcr hour per cm2. This amounts to about 0.5 mglcm2iday. By
chance, this is almost exactly the amount of nitroglycerine that can be induced to diffuse
through the skin. Nitroglycerine is a liquid at room temperature. It also has the requisite
physicochemical properties to place it at the high end of the permeability scale. Since it has
these ideal properties for transdermal delivery, it would appear that a flux of one half mg per
day per cm2 should be near the upper limit of attainable flux. At the lower end of the
permeability coefficient scale, the computed daily delivery is, for all practical purposes,
negligible unless permeation is from very high concentrations of a drug.3 Thus, u smull
permeability coeficient sets the second physical limitation on trunsdermal delivery.
Equation 5 indicates that a second general factor determining flux is the concentration
of the drug in the medium of application. As pointed out, the limiting mass transfer rate is seen
when permeation is from a saturated solution. In general, supersaturation leads to a metastable
situation and cannot be relied upon. In order to use equation 5 , one must know the solubility
of the drug in the medium that was used in the assessment of the permeability coefficient.
Almost invariably the solubility of interest will be the solubility in water, as the vast majority
of published permeability coefficients for the skin have been determined using aqueous media.
’Note that only relatively hydrophobic drugs have permeability coefficients as large as 1 x lo-’
cdhour, and these rarely have solubilities in water as high as 1 mgiml.
3However,large, polyfunctional, polar compounds tend to be highly crystalline and, consequently, none
too soluble in any phase.
178
Rynn and Stewart
Consequently, any scheme for estimating permeability coefficients will of necessity be
referenced to data generated in watery media.
Several completely separate factors have a bearing on solubility, including the ease of
dissociation of a molecule from its crystal and the degree of interaction of dissolved solute
molecules with the solvent. High-melting, hard crystalline materials with large enthalpies of
fusion are predictably less soluble than soft, low-melting crystals, everything else being equal.
It turns out that hydrophobic molecules have low intracrystalline cohesiveness and, according
to this factor, tend to be relatively soluble. However, the net negative free energy of solution
of hydrophobic molecules in water is small, meaning such molecules have low aqueous
solubilities. On the other hand, hydrophobic crystalline organics tend to be highly soluble in
nonpolar media, including the conduit phases of the skin. As a consequence, they have large
permeability coefficients, which tend to more than compensate for their low water
e ~For. this
~ reason, high fluxes are seen with moderately hydrophobic drugs.
Molecules containing multiple hydrogen bonding centers andlor strong dipoles are
generally high-melting because of strong intracrystalline self-association. While such molecules react strongly with water, imparting a degree of solubility in this solvent, they do not
dissolve substantially in organic phases, because of the exaggerated cohesive energy difference
between polar solute and nonpolar solvent. Consequently, partitioning into the lipoidal conduit
phases of the skin is minimal, and mass transfer rates are low even from their saturated
solutions. It is clear then that low solubility is the third physical obstacle to transdermal
delivery.
Transdermal Delivery Rate Projection
The above discussion indicates that the transdermal delivery rate of a drug can be
forecast providing one knows the drug’s solubility in and permeability coefficient from a given
solvent medium. It further suggests that the logical solvent medium on which to base such
projections is water. One can often find the aqueous solubilities of drugs in compendia or the
research literature. When the solubility of a compound of interest is not available, it is a
reasonably straightforward task to determine it, generally a 1-day experiment, providing there
is an ample supply of drug. In doing so, one has to account for the ionization of weak
electrolytes, so that the correct solubility of the drug can be assigned to the solid state
representing the nonelectrolyte form, for it will be the partition coefficient of the free form that
will be used to estimate a permeability coefficient. In principle, one also should be able to
es by one of several empirical methods. However, it has been our experience
that solubilities determined by the best of such methods, for instance, those based on the
melting points of drugs and their octanol/water partition coefficients [Yalkowsky, 19811, are
not sufficiently reliable for predictive purposes.
Permeability coefficients are neither prevalent in the literature nor so easily determined;
there is no immediately obvious means of estimating them that does not involve permeation
experiments. However, on careful examination of the literature, one finds prediction-useful
relationships between the permeability coefficients of drugs and their partitioning tendencies.
Flynn [ 1985) has discussed how the permeability coefficients of test compounds through
hairless mouse skin relate to the ethedwater partition coefficients of the compounds. The
reported data are displayed in Figure 3. It can be seen that compounds having molecular
weights greater than 200 daltons all seem to lie on a sigmoidal line stretching over ten
logarithmic orders of partition coefficient. The upper limit on the values is slightly above lo-*
40ne can view this another way. Although the solubility of a hydrophobic compound in water is low, the
product of this solubility and the partition coefficient into the skin is likely to be large, meaning that the
drug’s solubility in the surface structure of the skin is large. This goes hand in hand with the presumed
permeability coefficient.
Percutaneous Drug Penetration
179
O T
I-
z
w
2
LL
-I--
2
9
U
W
0
3
J,
H C - 2 1 ESTERS
I-
21
-3--
m
*GLvCERoL
U
W
U
’
dW
>.
2
.
4’
ALKAROLS
0 -2--
6
u
p6
-4--
--
UREA
W
*-
a
2
ara A
0 -5--
s
1
-6
HYDROCORTl<ONE
*.
Bra A 5 ESTEflS
-
8
-5
NEOMICIN
B
THIOUREA
GLUCOSE
I
-5
I
-4
I
-3
I
-2
I
I
1
I
I
I
I
-1
0
1
2
3
4
LOG ETHERlWATER PARTITION COEFFICIENT
Fig. 3. Plot showing how the permeability coefficients of a number of test compounds and drugs
through hairless mouse skin relate to their etheriwater partition coefficients. For compounds of molecular
weight > 200 daltons, all data fall on or reasonably close to a sigmoidal line spanning about ten log orders
of magnitude of partition coefficients (X-axis) and over three log orders of permeability coefficients
(Y-axis).
c d h o u r and is attained at etheriwater partition coefficients greater than 500 (log,, = 2.7).
Over the range of partition coefficients from 0.01 (loglo = -2) to 100 (log,, = +2), the
permeability coefficients increase by three logarithmic orders. Finally, compounds with log
(permeability coefficients) less than - 2 all fall on a line just above P =
cmihour. We
have found that these permeability coefficients fit to a similar general pattern on an
octanoVwater partition coefficient scale.
Permeability Analysis-Top
200 Drugs
To put the principles outlined above to a stringent test, it was decided to apply them to
an evaluation of the transdermal delivery potentials of the drugs in the top 200 new prescription
drug products of 1985’. The top 200 list is actually one of products, not chemical entities, and
is full of entity redundancy. Moreover, it was found that the necessary information to make
projections was available for only 22 (= 15%) of the actual chemical entities on the list.
Twenty-two compounds are nevertheless a sufficient cache of drugs to give fair assessment of
the utility of any estimative approach.
Establishing the daily transdermal drug requirement was the first step taken toward
estimating the transdermal potentials of the indicated drugs. It was reasoned that this would be
equal to the total 24-hour oral dose adjusted downward for oral first-pass loss. Oral dosages
were drawn from standard compendia [Huff, 1986; Boyd, 19811. First-pass information was
mainly found in Goodman and Gilman’s classic pharmacology text [Goodman et al., 19851.
No data on oral first-pass extraction could be found for 8 of the 22 compounds listed in Table
I . No attempt was made to account for conceivable but relatively improbable metabolism in
the skin (see the final paragraphs of this essay). The formula for computing the dose was:
5A manuscript by B. H. Stewart, G . L. Mynn, N. Sheth, and S . D. Roy entitled Physiochemical
assessment of the transdermal delivery potentials of drugs is in preparation.
180
FlynnandStewart
TABLE 1. Selected Drugs From the Top 200 (1985 List) Used to Test the Method of Assessing
Transdermal Delivery Potentials
Chemical name
Acetylsalicylic acid
Allopurinol
Amoxicillin
Chlorpropamide
Clonidine
Codeine phosphate
Dexamethasone
Erythromycin
Erythromycin estolate
Estrone sulfate
Haloperidol
Hydorcortisone acetate
Lorazepam
Methyldopa
Nicotine
Nitroglycerine
Penicillin V
Piroxicam
Phenytoin
Tetracycline
Theophylline
Trimethoprim
Total daily
transepidermal
dose"."
Permeability
coefficient'
( c d h r X lo3)
Aqueous
solubilityb
(mgiml)
3.25
0.016
1.48
0.036
3.82
0.41
19.6
20.0
20.0
20.0
20.0
5.27
20.0
0.2
2.96
20.0
20.0
6.21
20.0
0.014
0.19
1.63
3.3
0.48
4.0
2.2
30.0
8.3
0.1
2.0
0.024
0.03
0.014
0.01
0.08
10
100.0
1.25
0.25
1,ooo
250
-
0.01
20
300
1,000
385
2,000
0.013
1.70
8.3
0.4
(md
Estimated 24-hr
transepidermal
deliverye
(mg/cm2)
700
225
0.2
120
4
350
350
1.25
2.0
200
4
150
20
2.5
750
0.0257
0.00018
0.142
0.0019
2.75
0.082
0.047
0.960
0.0115
0.0144
0.0067
0.0013
0.0384
0.048
7.1
0.6
0.12
0.0015
0.0062
0.00057
0.038
0.0156
Patch
size
(cm2)
-
<1
-
85
-
87
-
104
2.8
4.2
-
-
"Permeability coefficients were estimated from the octanouwater partition coefficient using the formula
found in the text.
bAll values for the solubilities were obtained from the literature and were rounded to a comfortable
magnitude.
"The daily transdermal dose is equal to the daily oral dose X (100 - % extracted via first pass).
"No data on first-pass extraction could be found for the compounds with underlined values.
"See equation 8 in the text.
D2,
h, vansdcmal
- D24 hr, oral
x (100 - % extracted first-pass)
(7)
Each computed daily transdermal dose, D,, hr, transdemal, was rounded to the nearest whole
number; the doses are given in column 4 of Table I .
The literature was then scoured for the aqueous solubilities of the listed compounds. The
Merck Index [Windholz, 19831 was relied on heavily for these. The paucity of solubility
information on the compounds in the top 200 list limited the number of projections that could
be made. Very few reliable, experimental, aqueous solubilities were found. Solubilities
obtained from the literature are provided, in units of mg/ml, in columns of Table 1.
With the exceptions of nitroglycerine and clonidine, little if any skin permeation data
appear in the literature for the 22 test compounds. Thus all but two permeability coefficients
had to be estimated. Moreover, given the purpose of the exercise, the permeability coefficients
of nitroglycerine and clonidine were estimated exactly like the others so they might serve to
check the method. Permeability coefficients were also estimated for estradiol, fentanyl, and
scopolamine so that projected delivery rates could be compared with experimental data (Table
2). Therefore a means had to be devised to assign permeability coefficients to the drugs. It has
Percutaneous Drug Penetration
181
TABLE 2. Transdermal Delivery Calculations on Estradiol, Fentanyl, and Scopolamine
Chemical
name
Estradiol"
Fentanyl'
Scopolamine"
CS
(mg/ml)
P, actual
(cdhr) X lo3
P, estimated"
(cm/hr) X lo3
(mg/cm2/day)
(mg/cm2/day)
0.003
0.20
75.0
5.2
10.0
0.05
20.0
20.0
3.9
0.0004
0.0048
0.09
0.0014 (3.6)
0.0096 (2)
6.3 (70)
M24. aCNd
M24, estimatedb
"Estimated by the algorithm based on Koctanol,wa,er
partition coefficients outlined in the text of the article.
The partition coefficients were found in Hdnsch and Leo [19791.
parentheses: projected 24-hr delivery/experirnentally determined 24-hr delivery.
'Experimental data are from Michaels et al. [1975]. Yet to be published data on the permeability
coefficient and solubility of fentanyl from these laboratories are in very close agreement with the Michaels
et al. values.
dThe reported loglo (partition coefficient) of 1.26 on which this permeability coefficient value is based is
footnoted as "approximate." The value appears inconsistent with the high aqueous solubility of the
compound and its reported solubilities in apolar organic solvents such as benzene and pet ether. The
mineral loglo (oil/water) partition coefficient ib - 1.59 [Michaels et al., 19751; the loglo(heptane/water)
partition coefficient is -2.26 [Hansch and Leo, 19791.
been previously mentioned and is well established that permeability coefficients correlate well
with in vitro partition coefficients [Scheuplein, 1965; 1967; 1983; Michaels et al., 1975;
Elynn, 1985). Because of the existing extensive compilations of octanoliwater partition
coefficients [Hansch and Leo, 19791, the decision was made to scale the permeability
coefficients of the compounds to their Koctanovwaterpartition coefficients. Based on a
roughed-out plot of the permeability coefficients of certain compounds against their octanol/
water partition coefficients, the following algorithm was developed to estimate permeability
coefficients. For compounds with loglo (Koctanovwater)
< -2.301, a lower limit of I x
cmihour was assigned to the permeability coefficient. In this region the compounds were
considered polar enough to preferentially pass through the aqueous regime of the stratum
corneum LFlynn, 19851. For compounds with log 10 (Koctanol,water)values lying between
- 2.301 and
2.000, the permeability coefficients were determined from the empirically
derived relationship:
+
It is important that the steepness of this line and the nature of its placement on the octanoliwater
axis, which, as stated above, involved fitting the line through several experimental data points,
leaves considerable doubt about the accuracy, and therefore the usefulness, of permeability
coefficients estimated over this span of partition coefficients. For compounds with log,,
(Koctmovwater)values > 2.000, an upper limit permeability coefficient of 1 x lop2cmihour
was assigned. Compounds fitting into this partitioning range were considered sufficiently
nonpolar for their permeation rates to be controlled within aqueous strata of the epidermis and
dermis [Flynn, 19851.
In some instances, multiple values for the octanol/water partition coefficients of
compounds on the top 200 list have been reported; when there was a lack of agreement
between the values, the value that intuitively appeared to be the most reasonable was arbitrarily
chosen. Permeability coefficients were then computed (units of cmihour) and rounded to a
comfortable number; they are given in column 2 of Table 1 .
Using the solubilities and permeability coefficients, the cumulative amount of drug that
can be made to pass through a 1-cm2 area of skin from a saturated solution in a day, M24hr,
in milligrams was calculated from:
182
Flynn and Stewart
This equation will be recognized as an integrated form of equation 5. Column 5 of Table 1
gives the estimated 24-hour amounts. Only when within = 100-fold of the total daily
transdermal requirement (ratio of daily dose to M24hr--- 100) were patch sizes that would be
adequate to deliver the daily dose computed (last column in Table 1).
It can be seen from Table 1 that the method of estimation suggests that it is physically
possible to deliver clonidine and nitroglycerine by the transdermal method and, moreover, that
clonidine can be offered as a tiny transdermal device. Considering that transdermal products
are available for both drugs and that clonidine’s patch is indeed small (= 2.5 cm’), all three
of these projections have been borne out. Nicotine is also projected to be a good transdermal
candidate. Developments in several laboratories have demonstrated this to be true. However,
and more quantitatively, the projected daily deliveries of all three of these drugs appear
overestimated by three- to five-fold.
The analysis further suggests that dexamethasone, estrone sulfate, lorazepm, and
possibly haloperidol (ratio = 300) may be borderline candidates for the transdermal method.
Dexamethasone is not a transdermal candidate for systemic purposes, as it is a corticosteroid
well suited to oral administration. However, the drug is topically active. At this writing, there
are strong indications that estrone sulfate is being readied for transdermal use. Lorazepam also
seems to have genuine transdermal possibilities, but the same can be said for other highly
potent benzodiazepines. Whether haloperidol is a true candidate has to be researched, but the
possibility cannot be discounted. No other drug in the list is even close to meeting the minimal
specifications for transdermal delivery. Some fail because of high daily doses, while others fall
short because of intrinsically low skin permeabilities.
The predictive approach can be further challenged by comparing other drugs with the top
200 list. Predicting the fluxes of drugs currently available or being developed for transdermal
administration provides an interesting test. Therefore, Koctanovwate,-detenined
permeability
coefficients were estimated for estradiol, fentanyl, and scopolamine [Michaels et al., 197516
These are found in Table 2 alongside the known values. The estimated permeability
coefficients for both estradiol and fentanyl are within factors of four of their reported values,
certainly an acceptable level of agreement, and consequently the Mz4 values, actual and
projected, are equally close because the same solubility was used for the M24,
and
MZ4,
calculations. The predictive method seems to break down with scopolamine,
however. The estimated permeability coefficient is 70 times the experimental one; consequently, the estimated daily flux is also 70 times too large. While the octanol/water partition
coefficient for scopolamine is suspect (see footnotes to Table 2), we suspect the real problem
with the prediction for scopolamine results from a discrepancy in the relationship between
octanol/water partitioning and skin partitioning. In other words, the fundamental platform
upon which the predictive method is based seems itself to be weakening. The permeability is
overstated by over an order of magnitude. Since the error of estimation overstates rather th‘an
understates the potential of scopolamine, the compound would not be rejected out of hand as
a transdermal candidate, however. We suspect further breakdowns as this might occur with
highly water-soluble compounds that are also highly soluble in oils of intermediate polarity,
for it is for just such compounds that the correlation between octanoliwater partitioning and
skidwater partitioning is likely to languish.
We are currently compiling a broader data base to tighten the estimative method and
reveal where the method might fall short of expectation. The tendency to overestimate delivery
at the upper range of partition coefficient suggests that the upper limit value should be cut,
6Also including date for fentanyl from unpublished research done in these labs
Percutaneous Drug Penetration
183
possibly by a much as four-fold (to 5 X
cmihour). As more data are gathered, the best
placement for the line at intermediate polarity will become more evident. It almost doesn’t
matter where the lower limit line is set, as only in the rarest of circumstances will a drug in
the high polarity regime make the grade transdermally.
Frankly, we are encouraged by two aspects of the above analysis, first, that the
transdermal delivery potentials of so many compounds can be identified simply from physical
parameters that are typically gathered during preformulation exercises, and second, that no
good candidates “fell through the cracks. ” Presently, the limiting factors in making
predictions are the lack of availability of aqueous solubilities of the drugs and the disparities
in reported partition coefficients. Although we have placed the assessment of transdermal
delivery potential in a tighter quantitative framework than ever before, in reality we offer
nothing totally new concerning the factors that make transdermal delivery a possibility [Shaw
and Chandrasekaran, 1978; Chien, 1982; Cleary, 1983; Karim, 1983; Wester, 1985; Guy and
Hadgraft, 1985; Good, 1986; Walters, 19861. It takes potent drugs! Consider that the identified
transdermal candidate with the highest dose of those considered, nicotine, only requires an
estimated 20 mg to be delivered over the course of a day; this is high, as many, if not most,
candidates for the method of delivery, proven or suspected, have daily doses that are measured
in whole micrograms, not milligrams. Note also that all the promising candidates had
permeability coefficients at or near the established limiting value, meaning that all are
lipophilic. The idea that lipophilic drugs are the most likely drugs to “make it” isn’t fresh. But
why this dependency on lipophilicity? Here we point out that the most productive routes
through (or around) the principal barrier layer of skin, the stratum corneum, are lipoidal, which
favors solution of such species. This fact alone ties the opening discussion to the predictive
method. Moreover, lipoidal compounds exhibit low molecular cohesiveness, including
intracrystalline cohesiveness. Therefore they melt at lower temperatures and consume less total
energy in melting per mole than do polar organics. This makes them generally more soluble,
although obviously not so in water. This adds to the suitability of lipophilic drugs for
transdermal delivery, as they are not only more relatively soluble but also more absolutely
soluble in the skin’s critical diffusion conducting phases. The absolute solubility issue and its
relationship to low molecular cohesiveness ties in other parameters, like the melting points of
drugs, that have been used in a priori estimations of transdermal delivery potentials.
BIOLOGICAL CONSIDERATIONS IN TRANSDERMAL DELIVERY
lrritancy and Allergy Issues
Even when a drug has the requisite potency and permeability for transdermal delivery,
one is not sure the drug will make it to the marketplace in transdermal form because toxic
reactions within the skin resulting from application of the transdermal delivery system can still
scuttle the development project. Local imtancy associated with the drug andlor its delivery
system is a prime concern. The potential of any drug to induce an allergic response in the skin
is also worrisome. Both factors have caused a good deal of grief and disappointment along the
transdermal development path, often at the 1 1th hour when the highest price is paid for setback
and failure. Part of the problem is that human irritant responses, allergic sensitivities, and
photosensitivities are not actually predictable. Standard toxicity tests in animals are done in the
course of development of a transdermal device of course, but, unfortunately, such tests have
a poor record of extrapolation to the human condition. When no problems are encountered in
the animal models, the project team may proceed with a drug’s development falsely confident
and without further questioning. That the drug in question may already be in use by the oral
route without a significant incidence of skin toxicity only adds to the false sense of security.
It then comes as a shock when the exhaustively physically tested and fully developed drug
delivery system is not well tolerated in the clinic. Clearly, irritancy and allergy set a most
stringent constraint on transdermal delivery. Because of the stakes involved, it behooves the
184
FlynnandStewart
development team to obtain an early, direct reading of the human skin toxicity of a drug. This
requires strategic arranging of the sequencing of development events to bring human skin
tolerance testing into the foreground.
It should be pointed out that there is a straightforward reason why drugs that have clean
dermatotoxicity records when used orally and parenterally can run into toxicity problems when
topically deployed. Transdermal delivery leads to drug accumulations in the skin directly
beneath a patch that are many times higher than are ever achieved in the skin through systemic
administration of that drug. Imtancy (particularly) and allergic responses are known to be
concentration-dependent phenomena; topical administration has the potential of raising the
tissue concentrations of a drug (or other component of the transdermal system) above the
threshold for response. While it appears possible to modulate irritant responses, for instance
by designing a larger patch that delivers less drug per unit area, it seems that once the skin of
a patient is sensitized to a drug, that particular drug is forever contraindicated by the topical
route for that patient.
Metabolism During Transit Through the Skin
Another biological factor that can affect transdermal delivery is metabolism of the drug
in the skin coincident with its diffusive passage across the skin. It is conceivable that little
intact drug might reach the systemic circulation. Thus local first-pass metabolism at least has
the potential to act as another constraint on transdermal delivery. However, all current
experience suggests that this is not a general problem. For one thing, the skin has far less
capacity to metabolize drugs than have either the liver or the gastrointestinal mucosa, the latter
of which routinely synthesizes glucuronides and other adducts as drugs are absorbed.
Moreover, it appears from early observations in these labs that human skin has less capacity
to enzymatically alter drugs than do the skins of the rat and mouse. In particular, human skin
seems to have a low esterase activity. Other enzyme activities of the skin, human or otherwise,
seem to be substrate-specific and purposeful, and unlikely to cause problems with an
exogenous compound,
Enzymatic inactivation of a transdermally administered drug in the skin need not be
dismissed out of hand because of its improbability, for one can easily experimentally establish
whether or not concern should exist over the issue. The presence or absence of metabolites can
be demonstrated in in vitro diffusion cell experiments. Such studies also suggest the extent of
metabolism that will occur as the drug is absorbed. Also, the drug can be incubated with skin
homogenates to check for metabolism in the presence of all relevant enzyme cofactors. If
metabolism is possible at all, it will normally show up in such experiments.
REFERENCES
Boyd, J . R. (edj: “Drug Facts andComparisons,” 1982 edition. St. Louis: Facts and Comparisons, 1981.
Chien, Y.W.: Transdermal controlled-release drug administration. In: “Novel Drug Delivery SystemsFundamentals, Development Concepts, Biomedical Assessments. New York: Marcel Dekker,
Inc., 1982, pp. 149-217.
Cleary, G. W.: Transdermal controlled release systems, In Langer, R. S . and Wise, D. L. (eds):
“Medical Applications of Controlled Release” (Vol. I). Boca Raton, FL:CRC Press, 1983, pp.
203-25 1.
Elias, P., Grayson, S., Lampe, M. A., Williams, M. L., and Brown, B. E.: The intercorneocyte space.
In: Marks, R. and Plewig, G. (eds): “Stratum Corneum.” New York: Springer-Verlag, 1983, pp.
53-67.
Flynn, G. L.: Mechanism of percutaneous absorption from physicochemical evidence. In Bronaugh, R.
L. and Maibdch, H. I. (eds): “Percutaneous Absorption.” New York: Marcel Dekker, Inc., 1985,
pp. 17-42.
Goldsmith, L. A.: The epidermal cell periphery. In Goldsmith, L. A . , (edj: “Biochemistry and
Physiology of the Skin” (Vol I) New York: Oxford University Press, 1983, pp. 184-199.
PercutaneousDrug Penetration
185
Good, W. R.: Transdermal delivery systems, Med. Device Diagn. Ind. Feb. 1986: 35-42, 1986.
Goodman, A. G . , Gilman, L. S . , Rall, T. W., and Murad, F.: “Goodman and Gilmdn’s The
Pharmaceutical Basis of Therapeutics” 7th edition. New York: Macmillan Publishing Co., 1985,
pp. 1668-1733.
Guy, R. H. and Hadgraft, J.: Transdermal drug delivery: The ground rules are emerging, Pharm. Int.,
May 1985112-116, 1985.
Hansch, C. and LEO, A,: “Substituent Constants for Correlation Analysis in Chemistry and Biology.”
New York: John Wiley and Sons, 1979.
Huff, B. B. (managing ed): “Physician’s Desk Reference.” Oradell, NJ: Medical Economics Co., Inc.,
1986.
Karim, A.: Transdermal delivery systems. In “Drug Delivery Systems, Proceedings of an International
Conference” (Special Issue of Pharmaceutical Technology). Springfield, R: Aster Publishing Co.,
1983, pp. 28-32.
Katz, M. and Poulsen, B. J.: Absorption of drugs through the skin. In Brodie, B. B. and Gillette, J . (eds):
“Handbook of Experimental Pharmacology, New Series” (Vol. 28). New York: Springer-Verlag,
1971, pp. 103-174.
Kligman, A. M.: Perspectives and problems in cutaneous gerontology, J . Invest. Dermatol. 73: 39-46,
1979.
Michaels, A. S . , Chandrasekaran, S . K., and Shaw, J. E.: Drug permeation through human skin: Theory
and in vitro experimental method, A.1.Ch.E.J. 21: 985-996, 1975.
Odland, G. F.: Structure of the skin. In Goldsmith, L. A. (ed): “Biochemistry and Physiology of the
Skin” (Vol. I). New York: Oxford University Press, 1983, pp. 3-63.
Plewig, G . , Scheuber, E., Reuter, B., and Wdidelich, W.: Thickness of corneocytes. In Marks, R. and
Plewig, G . (eds): “Stratum Corneum.” New York: Springer-Verlag, 1983, pp. 171-174.
Scheuplein, K. J.: Mechanism of percutaneous absorption I: Routes of penetration and the influence of
solubility, J. Invest. Dermatol. 45: 334-346, 1965.
Scheuplein, R. J.: Mechanism of percutaneous absorption 11: Transient diffusion and the relative
importance of various routes of skin penetration, J. Invest. Dermatol. 48: 79-88, 1967.
Scheuplein, R. J.: Percutaneous absorption after twenty-five years: Or “old wine in new wineskins.” J.
Invest. Dermatol. 67: 31-38, 1976.
Scheuplein, R.: Skin as a barrier. In Jarrett, A. (ed): “The Physiology and Pathophysiology of Skin (Vol.
5 ) . New York: Academic Press, 1978, pp. 1693-1730.
Scheuplein, K.J. and Brouaugh, R. L.: Percutaneous absorption. In Goldsmith, L. A. (ed): “Biochemistry and Physiology of the Skin (Vol. 11) New York: Oxford University Press, 1983, pp.
1255-1295.
Shaw, J. E. and Chandrasekaran, R: Controlled topical delivery of drugs for systemic action, Drug
Metab. Rev. 8: 223-233, 1978.
Steinert, P. M.: Epidermal keratin: Filaments and matrix. In Marks, R. and Plewig, G . (eds): “Stratum
Coreneum.” New York: Springer-Verlag, 1983, pp. 25-38.
Walters, K. A.: Percutaneous absorption and transdermal delivery, Pharm. Technol. March 1986: 30,
32, 34, 36, 40, and 42, 1986.
Wester, R. C.: In vivo skin and nitroglycerine transdermal delivery, In Bronaugh. R. L. and Maibach,
H. I. (eds): “Percutaneous Absorption.” New York: Marcel Dekker, Inc., 1985, pp. 541-546.
Windholz, M. (ed): “The Merck Index,” 10th edition. Rahway, NJ: Merck and Company, Tnc., 1983.
Yalkowsky, S. H.: Solubility and solubilization of nonelectrolytcs. In Yalkowski, S . H. (ed):
“Techniques of Solubilization of Drugs.” New York: Marcel Dekker, 1981, pp. 1-14.